criteria, policy, procedures and division of professional tasks and responsibilities, are laid down in local and international, monodisciplinary and multidisciplinary agreements, defining women’s and children’s needs for healthcare services accordingly [26,51,59,84,85,115,120,122,132–136]. Timing is considered a highly important feature of the detection, assessment and decision making process. Generally, risk selection is addressed in two ways; the time until risk detection [28,31,137–139] and the time between risk detection and care provision [30,31,66,101,140,141]. It is emphasized that prolonged time between risk detection and care provision can cause delay, leading to preventable morbidity and mortality. Risk selection as a tool: Ensuring safety The majority of the papers we reviewed perceived risk selection as a tool to ensure safe care, the shared notion being that adequate risk selection results in safe care. Risk selection as a tool to ensure safe care is regulated by designated bodies [30,75,104,139,142]. The quality of risk selection is considered measurable, reflected by care outcomes. Authors assess the quality of risk selection predominantly by short-term and quantitative outcome measures, most frequently using referral rates, intervention rates and morbidity and mortality rates. The premise is that high quality risk selection results in low morbidity and mortality rates, and cases of maternal and perinatal morbidity and mortality that occur in specialist care reflect providers’ ability to screen for risks, to make a correct diagnosis, and to refer timely [30,31,62,101,140,143–148]. For example, Ferrazzi and colleagues [85] reflected on the results of their study on the outcomes of midwife-led labour in low-risk women: “as expected, maternal outcomes, such as mode of delivery, episiotomy and PPH, were significantly higher in women with compared to those without emerging risks identified by midwives. On the other hand, fetal outcomes were not significantly different between the two groups. This might be interpreted as a consequence of the quality of midwives’ assessment of risk during labor, which allowed for early diagnosis and prompt treatment of incoming complications”. While most studies focus on risk selection as a tool to ensure safe care, the standards for optimal risk selection and the optimality threshold remain unclear. Care outcomes are interpreted through cross-setting comparison, using a variety of reference points, including population, location and practice. For example, Rowe and colleagues [131] compared different types of maternity units, Fullerton and colleagues [149] compared local and national outcomes, Law and colleagues [150] compared midwife managed and obstetrician managed care, Romijn and colleagues [117] compared primary care midwives, clinical midwives and obstetricians, and Blondel and colleagues [151] compared countries. Furthermore, for measuring the quality of risk selection, quality is defined in various ways. A clear example is the diverse use of referral rates as a quality measure. According to Blix and colleague [65], “transfers should not be regarded as an adverse outcome, and are not necessarily indicators of quality of care”. Across the papers, high referral rates are perceived as indicators for effective risk selection [127,152] as well as failing risk selection [17,25,132,133]. Low referral rates are perceived in the same way; as indicators for effective risk selection [127] and failing risk selection [122,153]. Fourteen of the 210 papers use women’s experience as outcome of risk selection [61,124–126,128–130,154– 160] and one paper included partners’ experience as quality indicator for risk selection [161]. Overarching themes: Regulation, provider centred focus and avoiding underuse We found that the three dimensions of risk selection—an organisational measure to allocate resources, a practice to detect and assess risk and to make decisions about the delivery of care, ONE Risk selection in maternal and newborn care ONE | June 8, 2020 8 / 22 76 3 CHAPTER 3